Monitoring Particulate Matter in India: Recent Trends and Future Outlook

Air Quality, Atmosphere & Health https://doi.org/10.1007/s11869-018-0629-6

Monitoring particulate matter in India: recent trends and future outlook

Pallavi Pant 1 & RajM.Lal2 & Sarath K. Guttikunda3,4 & Armistead G. Russell2 & Ajay S. Nagpure5 & Anu Ramaswami5 & Richard E. Peltier1

Received: 25 May 2018 /Accepted: 28 September 2018 # Springer Nature B.V. 2018

Abstract Air quality remains a significant environmental health challenge in India, and large sections of the population live in areas with poor ambient air quality. This article presents a summary of the regulatory monitoring landscape in India, and includes a discussion on measurement methods and other available government data on air pollution. Coarse particulate matter (PM10) concentration data from the national regulatory monitoring network for 12 years (2004–2015) were systematically analyzed to determine broad trends. Less than 1% of all PM10 measurements (11 out of 4789) were found to meet the annual average WHO 3 Air Quality Guideline (20 μg/m ), while 19% of the locations were in compliance with the Indian air quality standards for PM10 (60 μg/m3). Further efforts are necessary to improve measurement coverage and quality including the use of hybrid monitoring systems, harmonized approaches for sampling and data analysis, and easier data accessibility.

Keywords India . Air quality monitoring . Air pollution policy . WHO Air Quality Guidelines . PM10 . PM2.5

Introduction the global population lives in areas that exceed the World Health Organization (WHO) Air Quality Guidelines, particu- Particulate matter (PM) has been identified as one of the most larly in low- and middle-income countries (van Donkelaar critical environmental risks globally (Brauer et al. 2015)and et al. 2015). PM also has climate effects including its direct poor air quality (AQ) including exposure to PM has been role in light scattering and absorption of sunlight and thermal associated with morbidity and mortality due to respiratory, radiation and the indirect role in acting as cloud condensation cardiovascular, and cerebrovascular diseases (Dockery et al. nuclei (CCN) (Ramanathan and Carmichael 2008), and im- 1993;Dominicietal.2006; Pope and Dockery 2006; Brook pacts visibility and general well-being (Levinson 2012; Singh et al. 2010;Thurstonetal.2016). A significant proportion of et al. 2017; Zhang et al. 2017).

Electronic supplementary material The online version of this article (https://doi.org/10.1007/s11869-018-0629-6) contains supplementary material, which is available to authorized users.

* Pallavi Pant Richard E. Peltier [email protected] [email protected]

Raj M. Lal 1 Department of Environmental Health Sciences, University of [email protected] Massachusetts, 686 North Pleasant St, Amherst, MA 01003, USA Sarath K. Guttikunda 2 School of Civil and Environmental Engineering, Georgia Institute of [email protected] Technology, Atlanta, GA, USA Armistead G. Russell 3 Division of Atmospheric Sciences, Desert Research Institute, 225 [email protected] Raggio Parkway, Reno, NV 89512, USA Ajay S. Nagpure 4 UrbanEmissions.Info, New Delhi, India [email protected] 5 Center for Science, Technology, and Environmental Policy, Hubert Anu Ramaswami H. Humphrey School of Public Affairs, University of Minnesota, [email protected] Minneapolis, MN, USA Air Qual Atmos Health

PM2.5 (particulate matter with an aerodynamic diameter matter, i.e., PM with aerodynamic diameter less than 10 μm; now less than 2.5 μm) is the specific fraction of the PM that is typically referred to as PM10), for residential, industrial, and sen- linked with human health impacts, and is often used as an sitive zones. The standards were revised in 2009, including the indicator for air quality within a community, though PM10 promulgation of a uniform standard for PM2.5 concentration (particulate matter with an aerodynamic diameter less than across all location types (http://cpcb.nic.in/National_Ambient_ 10 μm) is also found to have adverse health outcomes Air_Quality_Standards.php)(seeTable1). The standards for 3 3 (Ostro et al. 1999;HEI2011; Stieb et al. 2012;Shahetal. PM2.5,both24h(60μg/m ) and annual (40 μg/m ), are higher 2013). In India, 99.9% of the country’spopulationresidesin than the WHO Air Quality Guidelines (see Table 1), as well as areas that exceed the WHO Air Quality Guideline of 10 μg/m3 standards in the USA and Europe. The following section discusses (annual average), and half of the population resides in areas the major monitoring programmes in the country (see Table 2). where the Indian National Ambient Air Quality Standard 3 (NAAQS) for PM2.5 (40 μg/m ) is exceeded (Greenstone National Air Quality Monitoring Programme et al. 2015; Venkataraman et al. 2018). Per latest Global Burden of Disease (GBD) estimates, the average population- The National Air Quality Monitoring Programme (NAMP) is 3 weighted PM2.5 concentration is 74.3 μg/m , and exposure to the flagship air quality monitoring programme of the outdoor (ambient) air pollution is linked to 133.5 deaths per Government of India. As of September 2018, four key pollut-

100,000 people per year (Cohen et al. 2017). Another study ants—sulfur dioxide (SO2), nitrogen dioxide (NO2), PM10 estimated that exposure to PM2.5 is linked to ~600,000 pre- /RSPM, and PM2.5—are monitored at 703 AQ stations across mature deaths per year in India (Chowdhury and Dey 2016); 307 cities and towns (http://cpcb.nic.in/monitoring-network-3/). the status of epidemiological research with respect to air pol- The program is managed by the CPCB in coordination with the lution is summarized in Gordon et al. (2018). Despite poor air State Pollution Control Boards (SPCBs) and UT (union territo- quality, only a limited amount of data are available on air ry) Pollution Control Committees (PCCs). Other national orga- pollution trends in Indian cities, and there is a significant nizations such as NEERI (National Environmental Engineering gap in our understanding of spatio-temporal patterns of air Research Institute) and regional/local educational institutions pollutants at the local, regional, and national level. Previous are also involved in sample collection and analysis. studies have called for better coverage in terms of air pollution Most of these stations (residential and industrial) are locat- monitoring, as well as better QA/QC (quality assurance/ ed in urban areas, and the coverage in the rural areas is sparse quality control) to ensure high-quality data (Guttikunda et al. (Balakrishnan et al. 2014; Gordon et al. 2018). This is relevant 2014;Greenstoneetal.2015;Pantetal.2016). However, to to note since residential combustion emissions, typically asso- the best of our knowledge, there has been no concerted effort ciated with solid fuel use, are a key source of air pollution to summarize and evaluate existing regulatory AQ monitoring across the country (Balakrishnan et al. 2013; Venkataraman efforts and analyze associated datasets across India in the last et al. 2018). Typically, sampling is conducted twice a week, decade. Furthermore, there is a lack of comprehensive infor- with gas and PM samples collected for 24 h (4-h and 8-h mation on the types of monitoring networks, and associated periods respectively) (http://cpcb.nic.in/monitoring-network- data availability across the country. 3/). Outside of these 4- and 8-h time periods, no data is col- Considering the above issues and gaps, this manuscript lected at the manual stations based on publicly available in- reports on (1) regulatory PM2.5/PM10 data availability and formation. To ensure comparative sampling procedures across key trends throughout India, and (2) current status of the na- the NAMP stations, national guidelines have been set for tional monitoring networks, plans, and limitations in India. monitoring and include information regarding siting criteria, Results of our data analysis and the underlying data are avail- QA/QC procedures, measurement methods, and data able upon request. reporting (CPCB 2003; CPCB 2011). In addition to the monitoring sites under NAMP, several states including Maharashtra, Gujarat, Kerala, Odisha, Karnataka, Telangana, Ambient air quality monitoring in India and Andhra Pradesh conduct ambient AQ monitoring at additional sites under State Ambient Air Quality Monitoring Regulatory AQ monitoring, mandatedbytheAir(Preventionand Programme (SAMP). Control of Pollution) Act of 1981, is conducted in India under the The CPCB has also set up a network of continuous moni- aegis of the Central Pollution Control Board (CPCB) which toring stations (Continuous Automatic Air Quality operates under the Ministry of Environment, Forest and Climate Monitoring Station: CAAQMS) across major cities, and a

Change (MoEFCC) (Figs. S1, SI). The first set of standards for suite of pollutants including PM (PM2.5 and PM10), gases AQ in India was adopted in 1982 followed by revisions in 1994 (SO2,NO2,NH3, ground-level ozone (O3), CO [carbon mon- and 2009. Originally, there were separate standards for TSP (total oxide]), and BTEX (benzene, toluene, ethylbenzene, and xy- suspended particles) and RSPM (respirable suspended particulate lene) are measured continuously year-round. Currently, cities Air Qual Atmos Health

Table 1 Indian National Ambient Air Quality Standards (NAAQS) for fluorescence, GC-MS gas chromatography-mass spectrometry, FRM fed- PM and related components in India (TEOM tapered element oscillating eral reference method, FEM federal equivalent method, lpm liters per microbalance, AAS atomic absorption spectroscopy, ICP inductively minute) coupled plasma mass spectrometry, ED-XRF energy dispersive x-ray

Pollutant Unit Time duration NAAQS Standard Recommended measurement method WHO Air Quality Guidelines

3 PM2.5 μg/m Annual 24 h 40 PM2.5 Gravimetric measurement 10 60 Filter type: PTFE 47 mm 25

PM10 Annual 24 h 60 Volume: 16.7 lpm 20 100 Instrument: FRM or FEM equipment 50

PM10 Gravimetric measurement Filter type: GFA/ EPM 2000 Volume: 1132 lpm Instrument: High-volume sampler Real-time, continuous measurement TEOM/ β-attenuation Pb Annual 24 h 0.50 AAS/ICP on EPM 2000 or equivalent paper / 1.0 ED-XRF on PTFE filter Particulate BaP ng/m3 Annual 1.0 Solvent extraction followed by GC-MS analysis As 6.0 AAS/ICP on EPM 2000 or equivalent paper Ni 20.0

with a population of more than one million are prioritized for and SPCB, with a state and federal cost sharing financial CAAQM stations, and similar stations are expected to be set up arrangement. Data from the CAAQM stations is also used for across all state capitals and union territories (http://cpcb.nic.in/ determination of the air quality index (AQI), which is publicly Amb_AQ_Monitoring.pdf). As of September 2018, there are available online, via a smartphone app and historical data is 130+ CAAQM stations in more than 65 cities across India archived on the CPCB website. The AQI is calculated using (https://app.cpcbccr.com/ccr_docs/caaqms_list_All_India.pdf). the national AQI methodology, compiled and approved by the Most stations are operated as partnerships between the CPCB CPCB in 2014 (CPCB 2014).

Table 2 Summary of currently operational air quality networks in NAMP/SAMP/ SAFAR MAPAN US India CAAQMS Consulate

Purpose Regulatory x Research x x Other (e.g., public xxx awareness) Data Real-time x x x x Manual x Instrument type High-volume sampler(s) x BAM (or similar) x x x x

Pollutant PM10 xxx

PM2.5 xxxx Air quality Yes x x forecast No x x Open access to Real-time x x data Aggregate x x x None x Funding Government of India x x x Other x Air Qual Atmos Health

System of Air Quality and Weather Forecasting comprehensive. In most cities, the air quality monitoring and Research stations are being run in collaboration with local academic/ research institutions. Data generated as part of this network are The System of Air Quality and Weather Forecasting and expected to be used for evaluation of the IITM air quality Research (SAFAR) network (http://safar.tropmet.res.in/)was forecasting model, but very limited information is currently launched in Delhi in 2010 by the Indian Institute of Tropical available in the public domain. Meteorology (IITM, Pune) in collaboration with the Indian Meteorological Department (IMD) and the National Centre US Embassy/consulate monitoring stations for Medium Range Weather Forecasting (NCMWRF) and is supported by the Ministry of Earth Sciences (MoES). The As part of the AirNow program, the US Department of State program currently runs in Delhi (launched in 2010), Pune operates PM2.5 monitors in five major Indian cities—Delhi, (launched in 2013), Mumbai (launched in 2015), and Mumbai, Kolkata, Hyderabad, and Chennai—on the premises Ahmedabad (launched in 2017) with plans for expansion to of the Consulate/Embassy in each city. The data are publically Bengaluru, Kolkata, and Chennai. The network has ten air available in near real-time through the AirNow website. AQI quality monitoring stations (AQMS) in each city equipped values are also published on the website of the Consulate/ with real-time, continuous monitors for a range of pollutants Embassy; however, it is important to note that the reported including PM1,PM2.5,PM10,O3, nitrogen oxides (NOx), SO2, AQI value is based on the US air quality standards and uses black carbon (BC), methane (CH4), non-methane hydrocar- a different formula, and as such is not directly comparable to bons (NHMC), VOCs (BTEX) and mercury (Hg), and mete- the Indian AQI values. The instruments operate at the orological variables including solar radiation, rainfall, temper- embassy/consulate grounds, relatively isolated from heavy ature, relative humidity, wind speed, and direction. At some traffic or industrial sources, and in most cases they provide locations, ammonia (NH3) is also measured. Like the CPCB PM2.5 measurements equivalent to an urban location in the network, SAFAR also has a smartphone app, disseminating cities. calculated AQI values for the criteria pollutants (as a city Currently, data from the CAAQM stations are available average) and anticipated AQI for the 2 days. This information online (https://app.cpcbccr.com/ccr/#/caaqm-dashboard/ is also disseminated on real-time display boards in each city. caaqm-landing), but data access is often limited and, in some However, it is important to note that the AQI method utilized cases, unavailable. Data from the NAMP stations can also be under this program is different from the national approach downloaded through the Government of India open data approved by the CPCB (see Table S1) (Beig et al. 2015). platform (https://data.gov.in/). Most SPCB websites also AQ data from the network are embargoed for 3 years but are share ambient air quality data, but there are variations in the available for public use (upon request and justification, sub- type of data available. In most cases, monthly or annual mitted to IITM Pune) thereafter. averaged data are posted (most often as pdf files) on the SPCB websites. It must be noted that some states (e.g., Modelling Atmospheric Pollution and Networking Telangana, Tamil Nadu, Uttarakhand, Uttar Pradesh, and Maharashtra) have good data availability on their websites The Modelling Atmospheric Pollution and Networking (e.g., Uttarakhand- http://ueppcb.uk.gov.in/pages/display/95- (MAPAN) program is currently operational in ~20 cities air-quality-data; Maharashtra- http://mpcb.gov.in/envtdata/ across India and is supported by the MoES. Pollutants moni- envtair.php). Third-party organizations including several me- tored as part of this program include PM10 and PM2.5,O3,CO, dia outlets have also created data dashboards that allow the NOx,CO2, hydrocarbons, BC, and organic carbon (OC). PM public to access the data, and the raw data from the CAAQM monitoring is conducted using β-attenuation method (Met stations and the US Embassy/Consulate stations are archived One Instrument Model BAM-1020 in at least two locations by OpenAQ. Data from SAFAR and MAPAN networks are (Yadav et al. 2014; http://www.ncess.gov.in/research-groups/ not available in the public domain currently. atmospheric-processes-atp-group/laboratories/air-quality- monitoring-lab-aqml.html),butitisunclearifthesame instrument is used across all sites. Cities included in the Ambient PM10 concentrations across India network are Pune and Satara (Maharashtra), Thiruvananthapuram (Kerala), Bhubaneshwar (Odisha), Data collected as part of the NAMP network are released Chennai (Tamil Nadu), Jabalpur (Madhya Pradesh), monthly or annually, while the data from the CAAQM sta- Hyderabad (Telangana), Delhi (NCT of Delhi), Srinagar tions is accessible in near real-time (https://app.cpcbccr.com/ (Jammu and Kashmir), Udaipur (Rajasthan), Aizwal ccr/#/caaqm-dashboard/caaqm-landing). It is important to (Mizoram), Varanasi (Uttar Pradesh), Guwahati and Sonitpur note that no metadata or QA/QC data are made available, (Assam), and Patiala (Punjab), although this list is likely not either on the website or in the annual reports. Additionally, Air Qual Atmos Health for CAAQM stations, data is often missing for long periods of inconsistencies with monitor distribution (i.e., number of time often with no description for the reasons behind missing monitoring stations and location of monitors). However, an data. In terms of application, data collected as part of the assessment of PM10 monitoring data shows an increasing NAMP monitoring network is used for reporting annual trends trend for PM10 concentrations across the country in the last by the CPCB. Some trend analysis (Kandlikar 2007; few years (Yang et al. 2018). Additionally, between 1998 and

Guttikunda and Gurjar 2012; Gurjar et al. 2016) and epidemi- 2014, PM2.5 concentrations have steadily increased, and the ology studies (Balakrishnan et al. 2013;Majietal.2017)have number of districts (per 2011 Census) that exceed the annual 3 utilized the data, but there has been limited application of the average PM2.5 standard of 40 μ/m increased several fold (van data elsewhere. Recently, Pande et al. (2018) utilized the PM10 Donkelaar et al. 2015). Further, visibility data (a proxy for air data from NAMP stations to correct satellite data-derived es- pollution levels) from Indian cities shows an increase in bad timates (based on AOD) of PM10 at the country level. visibility days in the last decade (Hu et al. 2017). For this study, annual average PM10 data was retrieved Annual average PM10 concentrations were higher in from the NAMP stations from 2004 through 2015 from the Northern India (namely Rajasthan, Uttar Pradesh, Delhi, national data portal and the reports on the CPCB website. Jharkhand, and Punjab) compared to the Southern states

Monitoring of PM2.5 commenced only after the introduction (Fig. 2), consistent with other measurement, modeling, and sat- of the standard in 2009, and as a result, the data availability ellite findings (Nair et al. 2007; Dey et al. 2012; Saikawa et al. prior to this is comparatively limited. Although NAMP data 2017; Shi et al. 2018). Several studies have also reported on the was collected before 2004, national-level data was not avail- winter haze in the Indo-Gangetic plains in Northern India able. Measurements presented here are from 27 states and 5 (Gautam et al. 2007; Ram et al. 2016). The lowest annual aver- 3 UTs at 630 unique monitoring locations as of 2015 (see age PM10 concentration was 9 μg/m at Tekka Bench Ridge Table 2). A large number of sites exceed the annual PM10 (Shimla, Himachal Pradesh) in 2004 while the highest reported 3 standard (60 μg/m ), particularly in the northern and central annual average PM10 concentration at any sites was at Sarora parts of the country, and in some cases, the concentrations (Raipur, Chattisgarh) in 2011 with the annual concentration of were up to six times higher than the standard. 379 μg/m3 (annual average varied between 177 and 379 μg/m3 3 Only 11 of the 4789 (0.23%) unique PM10 annual average across sampling years, with an average of 273 μg/m ). The measurements meet the annual WHO Air Quality Guideline Tekka Bench Ridge station has been continuously active since (20 μg/m3) though 895 measurements (19%) met the annual 2004 and the annual average concentrations have ranged be- Indian NAAQS (60 μg/m3) (see Table 3). Over the 12-year tween 40 and 62 μg/m3 (average across the years is 42 μg/ period, only one site (Tekka Bench Ridge, Shimla, Himachal m3) except for 2004 (9 μg/m3) and 2005 (17 μg/m3). Overall, Pradesh) met the WHO AQ guideline in two separate years the highest annual average concentration averaged across all (2004 and 2005). Two hundred forty-six unique sites, mostly monitors in the state/UT in any year was 261 μg/m3 in Delhi located in southern and eastern India (see SI Fig. S2), met the (number of monitoring stations = 9) in 2010. Since 2008, Delhi Indian NAAQS in at least 1 year over the assessment period. has consistently been the most polluted state/UT although in the On the other hand, five sites (two in Chennai (Tamil Nadu) years prior, the northern state of Punjab recorded the highest and one each in Madurai (Tamil Nadu), Shillong (Meghalaya), PM10 levels. Several studies have reported high average con- and Mysore (Karnataka)) met the annual Indian standard each centrations for both PM10 and PM2.5 in Delhi (Chowdhury et al. of the 12 years that data is available. 2007; Guttikunda and Goel 2013;Pantetal.2015a, b;Tiwari

A summary of annual average PM10 concentrations shows et al. 2015; Sharma and Dikshit 2016;Majietal.2017;Yang significant variations across states, with higher concentrations et al. 2018). It is important to note that not all sources are the in the northern part of the country. There is little evidence of a same for both PM10 and PM2.5, and it is important to consider PM trend nationally. Over this time, the average number of the data with caution. In the Indian context, dust is a major stations in a state/UT has gone up from 8 to 19, while the source for PM10, and there is strong seasonality in the concen- PM10 concentration has remained more or less constant trations (Pande et al. 2018;Yangetal.2018). Studies in India (Fig. 1). Further, there were no notable trends within individ- have also indicated that the average PM2.5/PM10 ratio is be- ual states; it is difficult to assess state level trends as there are tween 0.5 and 0.8 (Kothai et al. 2011;Tiwarietal.2015).

3 3 Table 3 Number of annual average PM10 measurements attaining the Indian NAAQS (60 μg/m ) and WHO Guideline (20 μg/m )

2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2004 Total

WHOAirQualityGuideline00003000006211 Indian NAAQS 107 107 107 78 86 79 53 53 61 62 62 40 895 Total Sites 555 517 506 472 482 432 369 342 335 306 247 210 4789 Air Qual Atmos Health

Fig. 1 Average PM10 concentrations (panel A) and monitoring stations (panel B) based on NAMP data from 2004 to 2014

Observations from this dataset also align with available (dominant sources being residential biomass burning, dust, information on PM sources characteristics for the country fossil fuel and biomass burning, and waste burning (including crop residue)) (Pant and Harrison 2012; Guttikunda et al. 2014; Wiedinmyer et al. 2014;Pandeyetal.2014;Sadavarte and Venkataraman 2014;Karagulianetal.2015; Gargava and Rajagopalan 2015;Gurjaretal.2016). In a recent global

modeling assessment for energy-related sources of PM10,residential and industrial sectors were reported to be the largest contributors followed by power and transportation for India (Winijkul et al. 2015), and another study reported residential biomass fuel emissions to be the biggest contributor to ambi-

ent PM2.5 levels across India followed by combustion of agricultural residue (particularly in Northern India where in situ waste burning is a common practice) (Venkataraman et al. 2018). Non-exhaust emissions, including resuspended dust and brake, and tire emissions can also play a significant role

in PM10 emissions (Pant et al. 2015a, b; Nagpure et al. 2016). Year-on-year trend analysis was not conducted; if absolute

Fig. 2 Annual average PM10 concentrations at state/UT-level based on changes in PM10 concentrations are considered, there is a wide NAMP data from 2004 to 2015. [Note: Locations of monitors (industrial variation in trends across states. Overall, the number of mon- vs residential and other sites) tend to vary significantly from state to state. itoring stations has changed (in most cases, there has been an This figure should only be used for a general understanding of average pollution levels in urban India] increase in the number of monitoring stations) over time, and Air Qual Atmos Health it is expected that changes in demography and socio-economic sampling period does not capture the temporal variability of patterns would have impacted the emission patterns (Fig. 3). the pollutant. However, it is hard to make significant conclusions regarding Aerosol optical depth (AOD) measurements made on-board pollutant concentrations from these datasets since there has satellites offer another opportunity for estimation of PM concen- been a shift in the type of stations setup under NAMP during trations. Chowdhury and Dey (2016) estimated PM2.5 concen- the study period. While detailed information is not available, trations at the district level in India using daily columnar AOD there has been an increase in the number of residential/urban data from Multiangle Imaging SpectroRadiometer (MISR) and monitoring stations in the last decade. associated modeling exercises that combines the AOD data with There are limitations and uncertainties associated with this emission inventories and ground monitoring data using chemical dataset that have not been explicitly defined for each station; transport models. Based on this dataset, Kinnaur (Himachal 3 all measurements are gravimetric, and while there is a protocol Pradesh, 3.7 μg/m ) had the lowest annual average PM2.5 con- for sample collection and analysis, in the absence of QA/QC centration while the highest concentration was observed in Delhi data or information on laboratory inter-comparisons, it is hard (148 μg/m3). These observations are in broad agreement with the to interpret the quality of the data across the country. Further, reported trends for PM10 data from the NAMP network as most measurements occur over 24 h (8-h period) two to three discussed above. times a week, presumably during the day. This likely under- estimates the true PM10 concentration as nighttime concentrations often exceed daytime concentrations, and the 8-h Station density and coverage

The NAMP network currently has 700 stations and the network has come a long way since 1984 when it was first launched with seven stations in Northern India. The CPCB has provided detailed siting criteria for monitoring stations and included consid- erations such as size of the area, population, and pollutant sources (CPCB 2003). The guidelines also refer to a greater need for making measurements near traffic (diesel vehicles) and industries since these are major sources of PM. However, for a country with a population of ~1.2 billion, this amounts to one pollution monitoring station for every ~2 million people. Given the spatial heterogeneity of pollutant concentrations in India, in part due to mixed land-use (Tiwari 2002), the data should be improved (e.g., better coverage, clear metadata, and QA/QC) to better support pragmatic regulatory assessment or ep- idemiology studies. There is a wide variation in the number of monitoring stations across states often irrespective of the PM levels or population density (Fig. 4). Additionally, in several states including Manipur and Arunachal Pradesh in North-eastern India, station density is very poor and only 2–5 years of data is currently available. For comparison, there are 87 monitoring stations for

PM10 and 32 for PM2.5 in the Greater London region, UK (Mittal and Fuller 2017), a city of 9 million. To better characterize air pollution in Indian cities, there is a need to introduce more monitoring stations, especially in densely populated urban centers. As per the review of CPCB (IIM 2010), a total of 3000 stations were expected to be established in Class I (population of one million or more) and II (population of 50,000 to 99,999) cities and towns by 2022, and 100 CAAQM stations by 2012. However, there are currently 703 stations across the country, and 130 + CAAQM stations are operational as of September 2018. It is also important to consider that this dataset largely represents the air quality for urban and peri-urban India, since Fig. 3 Variability in PM 10 concentrations at NAMP stations in a 2004 and b 2015. [Note the higher concentrations in the Indo-Gangetic Plains, the NAMP network does not include monitoring stations in i.e., north India] rural areas (Balakrishnan et al. 2014). Given the significant Air Qual Atmos Health

Fig. 4 Average population density and average PM10 concentrations (in Dadra and Nagar Haveli, Daman and Diu, Tripura, Manipur, Sikkim, μg/m3) across states and the corresponding number of current NAMP Bihar, Telangana, Lakshadweep, Andaman and Nicobar Islands) monitoring stations [as of 2017] (following states/UTs are not included:

humanhealthburdenassociatedwithexposuretohousehold locations are meeting the 24-h NAAQS, since an 8-h sample air pollution, much of which occurs in rural areas, lack of data only captures a snapshot of the actual levels. Further, it is from rural areas could in fact be contributing to exposure likely that singular pollution episodes are either not accounted misclassification in large-scale modeling and epidemiological for, or have an undue influence on the measurements resulting studies. However, several new research projects are focusing in under- or over-reporting of PM10 concentrations. Manual on generating integrated ambient and indoor data on PM in monitoring also results in a delay in data collection, transmis- rural parts of the country (Balakrishnan et al. 2015; Tonne sion, and availability, although the increasing number of et al. 2017). CAAQMS are allowing data to be streamed to the CPCB Finally, from the perspective of air quality monitoring and website in near real-time overcoming some of the aforemen- assessment, it is important to establish regional monitoring tioned concerns. A case in point is Delhi, the capital city, stations for measurement of particle concentrations and com- which has the most extensive AQ monitoring network in the positional analysis at background levels. Data from such sta- country. Between 2004 and 2015, data capture for PM10 tions can be useful in assessment of long-term trends as well across ten sites (assuming 104 observations/year) varied be- as in source apportionment analyses, particularly with refer- tween 12 and 74% (Fig. 5). ence to understanding the contribution of regional and local A rigorous QA/QC procedure is vital to ensure that the sources. Additionally, such stations can provide valuable data results are comparable, reliable, and reproducible. The mea- on improving our understanding of regional transport of aero- surement and reporting protocols need to be consistent over sols. Currently, there are no regional monitoring stations with- time, and any biases due to instrument drift need to be in the regulatory (NAMP) network. accounted for in the analysis. A number of documents refer to QA/QC procedures for the NAMP monitoring stations Data quality (CPCB 2003; CPCB 2011); however, there is no information on data ratification and quality control for the NAMP moni- A majority of the air pollution monitoring stations across the toring sites, which suggests these data should be interpreted country are operated manually and it is unclear how the 8-h with caution. There is a Pollution Assessment, Monitoring and samples are used for regulatory purposes of assessing whether Survey (PAMS) program, but no information on data QA/QC Air Qual Atmos Health

Fig. 5 Monthly average PM10 concentration across ten stations in Delhi based on data retrieved for 2004 through 2015 [based on manual monitoring data reported by the CPCB] is available publicly. In case of SAFAR, a detailed QA/QC Future outlook for air quality data in India protocol is listed on the website, but neither the air quality data nor the QA/QC is publicly available. Similarly, there is no Air pollution remains a leading environmental concern in publicly accessible information for the MAPAN network. India, and a review of current monitoring and data manage- Traceable quality assurance requires clear and reproducible ment strategies indicates lack of consistency across different procedures which are implemented with scientific rigor and networks (e.g., differences in methodologies for estimation of need trained technicians and scientists to operate these net- AQI). Given the very high levels of PM in many parts of the works. In the Indian scenario, the issue of QA/QC is also tied country, India has also been in the international spotlight with to the technical capacity at the central and station pollution several cities listed among the worst in terms of air quality, and boards, and the zonal offices and laboratories. This includes increasingly, there is public pressure for action on air pollu- both the personnel and the infrastructure available to them for tion. Lack of centralized databases on air quality has rendered sampling and analysis. The CPCB review (IIM 2010) also it difficult to utilize available data for long-term trends analy- alluded to the issue of lack of technical capacity and limited sis studies, and in cases where the data is used, there is signif- opportunities for capacity building. Further, funding available icant uncertainty associated with final observations or out- to the central and state pollution control boards is often limited comes. Recent efforts such as open-access availability of and does not account for resources required for research and long-term datasets from NAMP stations is a welcome step analysis, thereby limiting rigorous data analysis and its direct and will undoubtedly spur and facilitate work on the topic. application in policymaking. Significant improvements have been made with respect to Air Qual Atmos Health air quality monitoring in India in the last two decades, and increase in the in situ urban populations in India (Bhagat efforts are underway to further strengthen the national moni- 2017). At the same time, recent trends indicate that the toring program, with a focus on setting up CAAQMS across net rural to urban migration constitute only a third of the major cities. Given the significant health burden associated total increase in urban populations (Bhagat 2017). Dey with exposure to PM in India, strengthening the regulatory et al. (2012) reported no correlation between annual air quality monitoring networks in India would provide greater PM2.5 (based on satellite retrieval) and population size support for both air quality management, higher order scien- but changes in the cities’ design and infrastructure tific inquiry, and epidemiological analysis activities. Here, we coupled with changes in demographics are likely to im- provide recommendations to further strengthen India’sair pact both pollution sources and exposures, and conse- quality monitoring networks: quently, the health impact estimates. As new monitoring stations are planned, such demographic changes can help [1] Improvement in data quality and data access: Third- to make the monitoring networks representative. party data evaluation, annual instrument inter- [3] Use of hybrid monitoring networks: Low-cost sensors comparisons and calibration can be used for data assess- offer an opportunity to generate high-resolution data at ment. In the UK, for example, all regulatory monitoring a lower cost, and with fewer deployment and access lim- data collected as part of the Automatic Urban and Rural itations (Snyder et al. 2013). A number of community Network (AURN) is ratified through Ricardo AEA, an projects have been launched worldwide for crowd- independent agency whereas in the USA, state and local funded air pollution measurements, and while such sen- monitoring agencies follow specifically developed stan- sors can be useful in designing citizen science projects dard operating procedures [SOPs] (often in consultation and generating novel data, there are still a number of with external contractors such as Sonoma Technology or uncertainties associated with the accuracy of measure- RTI International). The National Physical Laboratory of ments using low-cost sensors (Lewis and Edwards India (NPL) has been leading the charge in gas and PM 2016; Clements et al. 2017). So far, such monitors have calibrations (http://www.nplindia.in/gas-metrology). In not been proven to provide long-term, accurate data addition, it is important to collect and release metadata without systematic calibration (Lewis and Edwards (i.e., descriptive information about datasets) to ensure 2016; Rai et al. 2017). Efforts are underway to improve better cataloging and context-relevant usage of data. In precision among such sensors, and latest analyses are addition, information about the performance of the mon- supporting the case for deployment of well-designed itoring networks through key metrics such as annual data low-cost sensors for measurement of air pollution at the capture rate and instrument calibration and performance city level. If designed carefully, such networks can pro- evaluation should be archived and released. vide valuable data on spatial variability of pollutants and [2] Harmonized approaches for sampling and data analysis: help in identification of hyperlocal pollution hotspots. Appropriate siting of AQ monitoring stations can help Additionally, such instruments can be used as indicative meet the regulatory requirements of AQ monitoring and instruments to define and design more efficient regulato- can also be leveraged for exposure and health assessment ry monitoring networks. studies. Thus, it is crucial to setup monitoring stations In India, several independent low-cost sensor-based AQ based on where they are needed, and not merely where networks have been launched in recent years, and such they can be conveniently located. It would be useful to networks have been instrumental in improving public create a national guidance document on siting of AQ awareness. However, in the absence of reliable calibration, stations across a range of environments including urban and other QA/QC, it is unlikely that these data can be used (roadside, industrial, commercial, residential), rural, and for research or regulatory applications in the near future. regional monitoring areas. Regional monitoring stations [4] Utilizing alternative approaches to fill data gaps:One to measure background air pollution levels can be partic- avenue for improving our understanding of spatial and ularly helpful in determining the contribution of regional temporal patterns of PM is the use of satellite data. Data transport of aerosols. Several studies have reported sig- on aerosol optical depth (AOD) is typically collected via nificant contributions from non-local sources at the city satellite overpass measurements (e.g., MODIS- https:// as well regional level (Dey and Di Girolamo 2010). Data modis.gsfc.nasa.gov/) and can be used for AQ generated through the wide network of monitoring pro- applications such as forecasting, tracking pollution grams should also be utilized actively in designing strat- sources, and plumes, and as input/evaluation for AQ egies for pollution control, and for measuring the impact models (Duncan et al. 2014). Several recent studies have

of specific interventions. reported PM2.5 concentration estimates based on satellite One point of consideration for new monitoring sites is AOD retrieval and subsequent modeling using chemical the migration of people from rural to urban areas and an transport models (CTMs) (van Donkelaar et al. 2010; Air Qual Atmos Health

Dey et al. 2012; Brauer et al. 2015), and have contributed increased funding can help extend data collection. Such in improving our understanding of global pollution pat- data should not be limited to PM mass, but include spe- terns, especially in areas where ground monitoring is not ciation to help quantify the impacts of specific sources conducted, or historical data is not available. There are and can be immensely useful not only for regulatory still certain limitations, including inadequate bias- monitoring, but also for detailed health impact studies, correction based on local data, obtaining surface level source apportionment analyses, and assessment of policy measurements (since AOD is a columnar measurement) interventions. Similar recommendations have also been , and limited datasets (data retrieval is only possible on made elsewhere (Doraiswamy et al. 2017). For example, clear sky days). However, current results indicate that India is set to introduce Bharat Stage VI fuel standards satellite data can be very useful in generating annual (comparable to Euro VI) by 2020, and this is expected to

estimates for PM2.5, especially in areas with limited result in a decline in PM levels. With speciation data, it ground monitoring, both for long-term trend analysis, will be much easier to characterize changes in the emis- and for health impact assessments, but not for hour by sions’ profile after this national policy is introduced. At a hour or day to day measurements. MISR observations local level, speciation data, when collected over a long have additional information on PM physical properties term, can be helpful in measuring the impact of specific including aerosol size and shape. Additional satellite ob- policy interventions (e.g., odd-even vehicle policy in servations of gas-phase species, particularly PM precur- New Delhi) in terms of reduction of specific PM

sors (e.g., HCHO (formaldehyde), NO2,andSO2), can constituents. provide spatio-temporal information that can be related to PM levels and sources. A recent analysis by Pande Cities with a population of at least one million currently

et al. (2018) utilized AOD datasets to estimate PM10 account for ~ 30% of the country’s population, and as per concentrations and all-cause mortality associated with CPCB plans, the goal is to install CAAQMS (including all

PM10 exposures at the national level. criteria pollutants and meteorological parameters) in all state [5] Air Quality Forecasting and Early Warning Systems: capitals and cities with population of one million or more (see Given the high frequency of pollution episodes, it is im- the BNational Air Quality Monitoring Programme^ section). portant to have early warning systems which can be used A number of these cities are currently undertaking large infra- for alerting the public and broadcasting public health structure projects under the SMART City initiative or the advisories in urban and rural areas. In addition to gener- Urban Renewal Mission and have received federal aid. ating forecasts, designing public communications pro- Given the significant health burden associated with exposure grams to spread this information along with guidance to ambient air pollution, setting up, and operating CAAQM on what the public should (or should not do) through stations is a useful investment to get accurate, reliable, and apps, social media, and mass media (e.g., radio, TV and real-time information on air quality. Using CPCB siting print) in local languages would be useful. criteria, an average city of one million plus residents requires ~ 25 CAAQM stations, and if this number is extrapolated & SAFAR has been instrumental in providing advance across 60 cities, a total of ~1500 stations would be required information about the AQ levels in four cities, but (CPCB 2003). The average cost of a CAAQM station is ~INR similar systems based on high-quality data are needed one crore (~USD 0.15 million) with ~ 10% for annual main- across the country. Additionally, it is important to tenance costs (estimates based on figures reported for a station monitor data quality, and reliability of forecasts in to be commissioned in 2017 by the CPCB). This would re- the long term, which is currently lacking. quire an initial investment in the order of INR 3000 crores & In 2017, Urban Emissions, a private organization, (~USD 0.5 billion) due to capital and operational costs (INR launched an independent AQ forecasting system one crore per site for installation and INR one crore per site for where modeled hourly average concentration and a maintenance for 10 years). On top of this, costs associated 24-h moving average concentrations are simulated with infrastructure, personnel, and training need to be for the next 72 h, using the Weather Research and accounted for; and this can be estimated as an additional Forecasting (WRF) meteorological model and the INR 3000 crore (~USD 0.5 billion); to cover other miscella- GFS meteorological fields and the CAMx chemical neous costs, an additional 50% is added to this resulting in a transport modeling system (Skamarock et al. 2008). total of INR 7500 crores. These estimates indicate that the The data are available in the form of maps on an open- average cost of running the CAAQM station network in each access portal (http://www.indiaairquality.info). city over a 10-year period would amount to ~INR 12.5 crores per year (~USD 2 million/year) (see Table S2 for an estimate [6] Increase in expenditure and capacity building efforts:To of monitoring stations across Indian states). In comparison, further improve the country’s AQ monitoring network, the cost for the metro train infrastructure in major cities ranges Air Qual Atmos Health between INR 16,000 and 100,000 crores INR (~USD 2.5–15 matter air pollution and cardiovascular disease. Circulation 121: – billion). It is important to note that these estimates are based 2331 2378 Chowdhury S, Dey S (2016) Cause-specific premature death from ambi- on a broad calculation and are subject to change over time. ent PM2.5 exposure in India: estimate adjusted for baseline mortal- Finally, increased opportunities for collaboration between ity. Environ Int 91:283–290 the central and state pollution control boards, and local aca- Chowdhury Z, Zheng M, Schauer JJ, Sheesley RJ, Salmon LG, Cass GR, demic and research institutions, could also benefit pollution Russell AG (2007) Speciation of ambient fine organic carbon particles and source apportionment of PM2.5 in Indian cities. J Geophys mitigation efforts. Creating a system where research and Res Atmos 112 policymaking are better integrated and promoting research Clements AL, Griswold WG, Abhijit RS, Johnston JE, Herting MM, and development at the pollution control boards can help fill Thorson J, Collier-Oxandale A, Hannigan M (2017) Low-cost air the gaps. quality monitoring tools: from research to practice (a workshop summary). Sensors 17(11):2478 Cohen AJ, Brauer M, Burnett R, Anderson HR, Frostad J, Estep K, Funding information This work was financially supported by NSF Balakrishnan K, Brunekreef B, Dandona L, Dandona R, Feigin V, SRN# 1444745 & NSF PIRE Grant #1243535. Cesunica E. Ivey at Freedman G, Hubbell B, Jobling A, Kan H, Knibbs L, Liu Y,Martin UC-Riverside and James Hite at Georgia Tech assisted with data over- R, Morawska L, Pope CA, Shin H, Straif K, Shaddick G, Thomas sight. PM10 dataset used in this analysis is available via e-mail M, van Dingenen R, van Donkelaar A, Vos T, Murray CJL, ([email protected]). Forouzanfar MH (2017) Estimates and 25-year trends of the global burden of disease attributable to ambient air pollution: an analysis of Compliance with ethical standards data from the global burden of diseases study 2015. Lancet 389: 1907–1918 CPCB (2003) Guidelines for ambient air quality monitoring; series: Conflicts of interest Dr. Sarath Guttikunda is one of the founders of NAAQMS/ ... /2003-04. Central Pollution Control Board. https:// Urban Emissions, organization releasing air quality forecasts for Indian tinyurl.com/y8r54a3z. Accessed 12 Nov 2017 districts on an open-access platform. CPCB (2011) Guidelines for the measurement of ambient air pollutants volume-II. Central Pollution Control Board. https://tinyurl.com/ ycjs4h8j. Accessed 12 Nov 2017 References CPCB (2014) National Air Quality Index. Control of urban pollution series CUPS 82/2014–15. https://tinyurl.com/y78slqnq. Accessed 24 Jan 2018 Balakrishnan K, Ganguli B, Ghosh S, Sambandam S, Roy S, Chatterjee A Dey S, Di Girolamo L (2010) A climatology of aerosol optical and mi- (2013) A spatially disaggregated time-series analysis of the short- crophysical properties over the Indian subcontinent from nine years term effects of particulate matter exposure on mortality in Chennai, (2000–2008) of Multiangle Imaging SpectroRadiometer (MISR) India. Air Qual Atmos Health 6:111–121 data. J Geophys Res 115:D15204 Balakrishnan K, Cohen A, Smith K (2014) Addressing the burden of Dey S, Di Girolamo L, van Donkelaar A, Tripathi SN, Gupta T, Mohan M disease attributable to air pollution in India: the need to integrate (2012) Variability of outdoor fine particulate (PM2.5) concentration across household and ambient air pollution exposures. Environ – in the Indian subcontinent: a remote sensing approach. Remote Sens Health Perspect 122:A6 A7 Environ 127:153–161 Balakrishnan K, Sambandam S, Ramaswamy P, Ghosh S, Venkatesan V, Dockery DW, Pope CA, Xu X, Spengler JD, Ware JH, Fay ME, Ferris Thangavel G, Mukhopadhyay K, Johnson P, Paul S, Puttaswamy N, BG, Speizer FE (1993) An association between air pollution and Dhaliwal RS, Shukla DK, Team SRU-CAR (2015) Establishing mortality in six U.S. cities. N Engl J Med 329:1753–1759 integrated rural-urban cohorts to assess air pollution-related health Dominici F, Peng RD, Bell ML, Pham L, McDermott A, Zeger SL, Samet effects in pregnant women, children and adults in southern India: an JM (2006) Fine particulate air pollution and hospital admission for overview of objectives, design and methods in the Tamil Nadu air cardiovascular and respiratory diseases. JAMA 295:1127–1134 pollution and health effects (TAPHE) study. BMJ Open 5:e008090 Doraiswamy P, Jayanty RKM, Rao ST, Mohan M, Dey S, Ganguly D Beig G, Chate DM, Sahu SK, Parkhi NS, Srinivas R, Ali K, Ghude SD, et al. (2017) Combating air pollution in North India - the path for- Yadav S, Trimbake HK (2015) System of Air Quality Forecasting ward. Environmental Manager (EM) April 2017. http://pubs.awma. and Research (SAFAR - India). GAW Report No 217. https://www. org/flip/EM-Apr-2017/doraiswamy.pdf. Accessed 20 Jan 2018 wmo.int/pages/prog/arep/gaw/documents/GAWSymp2017P3Beig. Duncan BN, Prados AI, Lamsal LN, Liu Y, Streets DG, Gupta P, pdf. Accessed 12 Jan 2018. Migration and urban transition in India: Hilsenrath E, Kahn RA, Nielsen JE, Beyersdorf AJ, Burton SP, implications for development. United Nations Expert Group Fiore AM, Fishman J, Henze DK, Hostetler CA, Krotkov NA, Lee Meeting on sustainable cities, human mobility and international P, Lin M, Pawson S, Pfister G, Pickering KE, Pierce RB, Yoshida Y, migration Ziemba LD (2014) Satellite data of atmospheric pollution for U.S. Bhagat RB (2017), UN/POP/EGM/2017/3. https://tinyurl.com/ air quality applications: examples of applications, summary of data y82sb8zw. Accessed 18 Sep 2018 end-user resources, answers to FAQs, and common mistakes to Brauer M, Freedman G, Frostad J, van Donkelaar A, Martin RV,Dentener avoid. Atmos Environ 94:6472–66 F, Van Dingenen R, Estep K, Amini H, Apte JS, Balakrishnan K, Gargava P, Rajagopalan V (2015) Source apportionment studies in six Barregard L, Broday DM, Feigin V, Ghosh S, Hopke PK, Knibbs Indian cities: drawing broad inferences for urban PM10 reductions. LD, Kokubo Y, Liu Y, Ma S, Morawska L, Texcalac Sangrador JL, Air Qual Atmos Health (5):471–481 Shaddick G, Anderson HR, Vos T, Forouzanfar MH, Burnett RT, Gautam R, Hsu NC, Kafatos M, Tsay S (2007) Influences of winter haze Cohen A (2015) Ambient air pollution exposure estimation for the on fog/low cloud over the Indo-Gangetic Plains. J Geophys Res global burden of disease 2013. Environ Sci Technol 50(1):79–88 Atmos 112 Brook RD, Rajagopalan S, Pope CA, Brook JR, Bhatnagar A, Diez-Roux Gordon T, Balakrishnan K, Dey S, Rajagopalan S, Thornburg J, Thurston AV, Holguin F, Hong Y, Luepker RV, Mittleman MA, Peters A, G, Agrawal A, Collman G, Guleria R, Limaye S, Salvi S, Kilaru V, Siscovick D, Smith SC, Whitsel L, Kaufman JD (2010) Particulate Nadadur S (2018) Air pollution health research priorities for India: Air Qual Atmos Health

perspectives of the Indo-U.S. Communities of Researchers. Environ Pandey A, Sadavarte P, Rao AB, Venkataraman C (2014) Trends in multi- Int 119:100–108 pollutant emissions from a technology-linked inventory for India: II. Greenstone M, Nilekani J, Pande R, Ryan N, Sudarshan A, Sugathan A Residential, agricultural and informal industry sectors. Atmos (2015) Lower pollution, longer lives: Life expectancy gains if India Environ 99:341–352 reduced particulate matter pollution. Econ Polit Wkly L 40–46 Pant P, Harrison RM (2012) Critical review of receptor modelling for Gurjar BR, Ravindra K, Nagpure AS (2016) Air pollution trends over particulate matter: a case study of India. Atmos Environ 49:1–12 Indian megacities and their local-to-global implications. Atmos Pant P, Baker SJ, Shukla A, Maikawa C, Godri Pollitt KJ, Harrison RM Environ 142:475–495 (2015a) The PM10 fraction of road dust in the UK and India: char- Guttikunda SK, Goel R (2013) Health impacts of particulate pollution in a acterization, source profiles and oxidative potential. Sci Total megacity—Delhi, India. Environ Dev 6:8–20 Environ 530–531:445–452 Guttikunda SK, Gurjar BR (2012) Role of meteorology in seasonality of Pant P, Shukla A, Kohl SD, Chow JC, Watson JG, Harrison RM (2015b) air pollution in megacity Delhi, India. Environ Monit Assess 184: Characterization of ambient PM2.5 at a pollution hotspot in New 3199–3211 Delhi, India and inference of sources. Atmos Environ 109:178–189 Guttikunda SK, Goel R, Pant P (2014) Nature of air pollution, emission Pant P, Guttikunda SK, Peltier RE (2016) Exposure to particulate matter sources, and management in the Indian cities. Atmos Environ 95: in India: a synthesis of findings and future directions. Environ Res 501–510 147:480–496 HEI (2011) Public Health and Air Pollution in Asia (PAPA): coordinated Pope CA, Dockery DW (2006) Health effects of fine particulate air pol- studies of short-term exposure to air pollution and daily mortality in lution: lines that connect. J Air Waste Manage Assoc 56:709–742 two Indian cities. Research Report Number 157. https://www. Rai AC, Kumar P, Pilla F, Skouloudis AN, Di Sabatino S, Ratti C, Yasar healtheffects.org/system/files/India-157-IR.pdf Accessed 23 A, Rickerby D (2017) End-user perspective of low-cost sensors for Nov 2017 outdoor air pollution monitoring. Sci Total Environ 607–608:691– Hu Y, Yao L, Cheng Z, Wang Y (2017) Long-term atmospheric visibility 705 trends in megacities of China, India and the United States. Environ Ram K, Singh S, Sarin MM, Srivastava AK, Tripathi SN (2016) – Res 159:466 473 Variability in aerosol optical properties over an urban site, Kanpur, IIM (2010) Evaluation of Central Pollution Control Board (CPCB). IIM in the Indo-Gangetic Plain: a case study of haze and dust events. Lucknow. http://cpcbnicin/cpcbold/IIMLkopdf. Accessed 15 Atmos Res 174:52–61 Nov 2017 Ramanathan V, Carmichael G (2008) Global and regional climate chang- Kandlikar M (2007) Air pollution at a hotspot location in Delhi: detecting es due to black carbon. Nat Geosci 1:221–227 trends, seasonal cycles and oscillations. Atmos Environ 41:5934– Sadavarte P, Venkataraman C (2014) Trends in multi-pollutant emissions 5947 from a technology-linked inventory for India: I. industry and trans- Karagulian F, Belis CA, Dora CFC, Prüss-Ustün AM, Bonjour S, Adair- port sectors. Atmos Environ 99:353–364 Rohani H, Amann M (2015) Contributions to cities’ ambient partic- Saikawa E, Trail M, Zhong M, Wu Q, Young CL, Janssens-Maenhout G, ulate matter (PM): a systematic review of local source contributions Klimont Z, Wagner F, Kurokawa J, Nagpure AS, Gurjar BR (2017) at global level. Atmos Environ 120:475–483 Uncertainties in emissions estimates of greenhouse gases and air Kothai P, Saradhi IV, Pandit GG, Markwitz A, Puranik VD (2011) pollutants in India and their impacts on regional air quality. Chemical characterization and source identification of particulate Environ Res Lett 12:065002 matter at an urban site of Navi Mumbai, India. Aerosol Air Qual Res 11:560–569 Shah ASV, Langrish JP, Nair H, McAllister DA, Hunter AL, Donaldson Levinson A (2012) Valuing public goods using happiness data: the case K, Newby DE, Mills NL (2013) Global association of air pollution of air quality. J Public Econ 96:869–880 and heart failure: a systematic review and meta-analysis. Lancet – Lewis A, Edwards P (2016) Validate personal air-pollution sensors. 382:1039 1048 Nature 535:29–31 Sharma M, Dikshit O (2016) Comprehensive Study on Air Pollution and Maji S, Ahmed S, Siddiqui WA, Ghosh S (2017) Short term effects of Green House Gases (GHGs) in Delhi. http://delhi.gov.in/DoIT/ criteria air pollutants on daily mortality in Delhi, India. Atmos Environment/PDFs/Final_Report.pdf. Accessed 12 Dec 2017 Environ 150:210–219 Shi Y,Matsunaga T, Yamaguchi Y,Li Z, Gu X, Chen X (2018) Long-term Mittal L, Fuller G (2017) London Air Quality Network, Summary Report trends and spatial patterns of satellite-retrieved PM2.5 concentra- 2016, King’s College London. https://www.londonair.org.uk/ tions in South and Southeast Asia from 1999 to 2014. Sci Total – london/reports/2016_LAQN_Summary_Report.pdf. Accessed 12 Environ 615:177 186 Jan 2018 Singh A, Bloss WJ, Pope FD (2017) 60 years of UK visibility measure- Nagpure AS, Gurjar BR, Kumar V, Kumar P (2016) Estimation of ex- ments: impact of meteorology and atmospheric pollutants on visi- haust and non-exhaust gaseous, particulate matter and air toxics bility. Atmos Chem Phys 17:2085–2101 emissions from on-road vehicles in Delhi. Atmos Environ 127: Skamarock WC, Klemp JB, Dudhia J, Gill DO, Barker DM, Wang W, 118–124 Powers JG (2008) A Description of the Advanced Research WRF Nair VS, Moorthy KK, Alappattu DP, Kunhikrishnan PK, George S, Nair Version 3. NCAR Tech. Note NCAR/TN-475+STR PR, Babu SS, Abish B, Satheesh SK, Tripathi SN, Niranjan K, Snyder EG, Watkins TH, Solomon PA, Thoma ED, Williams RW, Hagler Madhavan BL, Srikant V, Dutt CBS, Badarinath KVS, Reddy RR GSW, Shelow D, Hindin DA, Kilaru VJ, Preuss PW (2013) The (2007) Wintertime aerosol characteristics over the Indo-Gangetic changing paradigm of air pollution monitoring. Environ Sci Plain (IGP): impacts of local boundary layer processes and long- Technol 47:11369–11377 range transport. J Geophys Res Atmos 112 Stieb DM, Chen L, Eshoul M, Judek S (2012) Ambient air pollution, birth Ostro BD, Hurley S, Lipsett MJ (1999) Air pollution and daily mortality weight and preterm birth: a systematic review and meta-analysis. – in the Coachella Valley, California: a study of PM10 dominated by Environ Res 117:100 111 coarse particles. Environ Res 81(3):231–238 Thurston GD, Burnett RT, Turner MC, Shi Y, Krewski D, Lall R, Ito K, Pande P, Dey S, Chowdhury S, Choudhary P, Ghosh S, Srivastava P, Jerrett M, Gapstur SM, Diver WR, Pope IC (2016) Ischemic heart Sengupta B (2018) Seasonal transition in PM10 exposure and asso- disease mortality and long-term exposure to source-related compo- ciated all-cause mortality risks in India. Environ Sci Technol 52(15): nents of U.S. fine particle air pollution. Environ Health Perspect 8756–8763 124:785–794 Air Qual Atmos Health

Tiwari G (2002) Urban transport priorities: meeting the challenge of Venkataraman C, Brauer M, Tibrewal K, Sadavarte P, Ma Q, Cohen A, socio-economic diversity in cities, a case study of Delhi, India. Chaliyakunnel S, Frostad J, Klimont Z, Martin RV, Millet DB, Cities 19:95–103 Phillip S, Walker K, Wang S (2018) Source influence on emission Tiwari S, Hopke PK, Pipal AS, Srivastava AK, Bisht DS, Tiwari S, Singh pathways and ambient PM2.5 pollution over India (2015–2050). AK, Soni VK, Attri SK (2015) Intra-urban variability of particulate Atmos Chem Phys 18:8017–8039 matter (PM2.5 and PM10) and its relationship with optical properties Wiedinmyer C, Yokelson RJ, Gullett BK (2014) Global emissions of of aerosols over Delhi, India. Atmos Res 166:223–232 trace gases, particulate matter, and hazardous air pollutants from Tonne C, Salmon M, Sanchez M, Sreekanth V, Bhogadi S, Sambandam open burning of domestic waste. Environ Sci Technol 48:9523– S, Balakrishnan K, Kinra S, Marshall JD (2017) Integrated assess- 9530 ment of exposure to PM2.5 in South India and its relation with Winijkul E, Yan F, Lu Z, Streets DG, Bond TC, Zhao Y (2015) Size- cardiovascular risk: design of the CHAI observational cohort study. resolved global emission inventory of primary particulate matter Int J Hyg Environ Health 220:1081–1088 from energy-related combustion sources. Atmos Environ 107:137– van Donkelaar A, Martin RV, Brauer M, Kahn R, Levy R, Verduzco C, 147 Villeneuve PJ (2010) Global estimates of ambient fine particulate Yang X, Jiang L, Zhao W, Xiong Q, Zhao W, Yan X (2018) Comparison matter concentrations from satellite-based aerosol optical depth: de- of ground-based PM and PM concentrations in China, India, – 2.5 10 velopment and application. Environ Health Perspect 118:847 855 and the US. Int J Environ Res Public Health 15:1382 van Donkelaar A, Martin RV, Brauer M, Boys BL (2015) Use of satellite Zhang X, Zhang X, Chen X (2017) Happiness in the air: how does a dirty observations for long-term exposure assessment of global concen- – sky affect mental health and subjective well-being? J Environ Econ trations of fine particulate matter. Environ Health Perspect 123:135 Manag 85:81–94 143